Repository navigation
fix(product_association): target_item rejects integer product ids #594
Description
What: ProductAssociation(target_item=...) rejects integer product ids, the most common kind. 10, np.int64(10) and [10] all raise TypeError: target_item must contain only str or float values. Checked on main @ c5f4b2a.
Source: product_association.py:163.
The exclusion looks accidental:
10.0is accepted, and filters correctly against an integerproduct_idcolumn.3.5is accepted, and returns an empty frame.- e2fcc3b ("Support str and float data types for target items") gives no reason for leaving out int.
- The only test of the check (
test_calc_association_target_item_invalid_type) passes a dict, not an int.
Expected: accept int and np.integer, alone or in a list. Related: #587 (annotation mismatches in this module).
Failing test (3 fail on main):
import numpy as np
import pandas as pd
import pytest
from openretailscience.analysis.product_association import ProductAssociation
from openretailscience.options import ColumnHelper
cols = ColumnHelper()
class TestProductAssociationIntegerTargetItem:
@pytest.fixture
def transactions_df(self) -> pd.DataFrame:
return pd.DataFrame(
{
cols.transaction_id: [1, 1, 2, 2, 3],
"product_id": [100234, 100877, 100234, 100951, 100877],
},
)
@pytest.mark.parametrize("target_item", [100234, np.int64(100234), [100234]])
def test_integer_target_item_filters_to_that_product(self, transactions_df, target_item):
result = ProductAssociation(
transactions_df,
value_col="product_id",
group_col=cols.transaction_id,
target_item=target_item,
).df
assert set(result["product_id_1"]) == {100234}
assert set(result["product_id_2"]) == {100877, 100951}
Activity